Dynamic Demand for New and Used Durable Goods Without Physical Depreciation: the Case of Japanese Video Games∗

Dynamic Demand for New and Used Durable Goods without Physical Depreciation: The Case of Japanese Video Games∗ Masakazu Ishiharay Andrew Ching Stern School of Business Rotman School of Management New York University University of Toronto This draft: April 21, 2016 ∗We have benefited from discussions with Sridhar Moorthy, Ron Borkovsky, Victor Aguirregabiria, Harikesh Nair, Avi Goldfarb, Susumu Imai, Nitin Mehta, Mengze Shi, Tanjim Hossain, David Soberman, Ig Horstmann, Andy Mitchell, Alon Eizenberg, Christos Genakos, Benjamin Shiller. We also thank seminar and conference participants at U of Toronto, Erasmus School of Economics, NYU, U of Michigan, U of Rochester, UT-Dallas, UBC, NUS, HKUST, Washington University-Saint Louis, OSU, UCLA, Columbia, Northwestern, UCSD, McMaster, UC Berkeley, Stanford, Chicago, 2009 Marketing Dynamics Conference, 2010/2012 IIOC, 2012 CKGSB Marketing Research Forum, 2012 NBER Summer Institute Economics of IT and Digitization Workshop, 2014 Society for Economic Dynamics Conference, and 2015 ZEW Conference for their helpful comments. All remaining errors are ours. yPlease direct all correspondence to Masakazu Ishihara, Stern School of Business, New York University. Email: mishi- [email protected]. Dynamic Demand for New and Used Durable Goods without Physical Depreciation: The Case of Japanese Video Games Abstract For information/digital products, the used goods market has been viewed as a threat by producers. However, it is not clear if this view is justified because the used goods market also provides owners with an opportunity to sell their products. To investigate the impact of the used goods market on new goods sales, we collect a unique data set from the Japanese video game market. Based on the data, we develop and estimate a new dynamic structural model of consumers' buying and selling decisions. The estimation results show that (i) the consumption value of owners depreciates much faster than that of potential buyers, and (ii) consumers are forward-looking but they discount the future more heavily than what the interest rate suggests. Using the estimates, we quantify the impact of eliminating the used game market on publishers' profits and consumer welfare. We find that this policy would increase publishers' profit by 2%, but reduce the consumer surplus by 19% if publishers do not adjust their prices. However, if they adjust prices optimally, it would increase the profit by 78%, and also increase the consumer surplus by 141% due to lower new prices. Overall, our results suggest that the elimination of the used good market improves the social welfare. Keywords: Information/Digital Products, Durable Goods, Used Goods Market, Demand Estimation, Dy- namic Programming, Transaction Costs, Satiation, Discount Factor, Bayesian Estimation. 1 Introduction The existence of used goods markets has been viewed as a serious problem by producers in information/digital product categories such as books, CDs/DVDs, and video games. They argue that the competition from used goods significantly lowers their profits and reduce the incentive to develop new products. For instance, book publishers and authors expressed their annoyance to Amazon over used books sold on its websites (Tedeschi 2004). Video game publishers in Japan attempted to kill off used video game retailing by suing used video game retailers (Hirayama 2006). Their main argument is that products like books and video games physically depreciate negligibly, but owners' consumption values can decline very quickly due to satiation. As a result, unlike products that physically depreciate more considerably (such as cars), producers of information/digital products may face competition from used goods that appear to be almost identical to new goods soon after the release of a new product. However, their argument focuses only on one aspect of used goods markets (substitution effect), and ignores the possibility that costs of buying and selling used goods could reduce the substitutability between new and used goods. Moreover, the existence of used goods markets provides consumers with a selling opportunity. If consumers are forward-looking and account for the future resale value when making a buying decision, the effective price consumers pay for a product will be lower than the actual price (resale effect). This feature implies that the existence of used goods markets could increase the sales of new goods. Thus, whether the existence of used goods markets hurts or benefits new-good producers is an empirical question, and the answer depends on which effect, substitution or resale effect, dominates. The question of whether used goods markets help or hurt new-good producers is not new and has been investigated in the automobile market (e.g., Esteban and Shum 2007, Chen et al. 2013, Schiraldi 2011), and in the housing market (e.g., Tanaka 2013). A general finding among these papers is that the elimination of used goods markets helps new-good producers. For example, Chen et al. (2013) find that opening the used good market lowers firms’ profits by 35%. However, it is not straightforward whether these findings will extend to markets such as books and video games, where the sales of new goods is highly concentrated 1 around the release period, and declines quickly afterwards. If the incentive for consumers to buy new goods in the release period is largely driven by the future resale opportunity, then the elimination of the used goods market could significantly reduce the initial sales of new copies, and new-good producers might not be able to recover its initial loss in subsequent periods even if cannibalization from used goods is eliminated. This paper contributes to the existing literature in two important dimensions. First, we assemble a new data set from the Japanese video game market, which includes weekly aggregate level data for 20 video game titles released in Japan between 2004 and 2008. The novel aspect of this data set is that, in addition to the sales and prices of new and used goods, which are main variables studied in previous works, it includes three new important variables: (i) resale values of used goods,1 (ii) quantities of used goods retailers purchased from consumers, and (iii) aggregate inventory level of used goods at retailers. This novel data set allows us to empirically capture important distinctions between video games and other typical durable goods such as cars and houses studied in previous works. One common feature assumed in most of the previous research (both theoretical and empirical) is that durability is measured as the quality deterioration rate, and it is common across buyers and sellers. This assumption will likely be violated in information/digital product categories such as CDs/DVDs and video games because for product owners, consumption values deteriorate mainly due to satiation (satiation-based deterioration); but for potential buyers, consumption values may deteriorate due to freshness of a product (freshness-based deterioration). Our new data on used game trading activities (weekly used-copy quantities demanded and supplied by consumers and associated weekly prices and resale values) help identify these two forms of consumption value deterioration without the need to make an assumption on the used-good market clearing condition. Second, based on the data set and institutional details about the video game market, we develop and estimate a new structural model of consumers' buying and selling decisions of durable goods which do not exhibit physical depreciation. To our knowledge, this is the first dynamic model of forward-looking consumers that incorporates all of the following features: (i) new and used goods buying decisions, (ii) used 1We define resale value as the amount consumers receive when they sell their used video games to retailers. In Japan, retailers usually set a take-it-or-leave-it resale value for each game, and consumers sell their games at that resale value. Negotiation is uncommon. 2 goods selling decision, (iii) consumer expectations about future prices of new and used goods, resale values of used goods, and inventory levels of used goods, (iv) costs of buying and selling used goods, (v) the impact of used goods availability on buying decisions, and (vi) deterioration of both owners' and potential buyers' consumption values. In our model, the expected discounted value of future payoffs from buying a product is determined by a dynamic consumer selling decision problem, which depends on the deterioration rate of owners' consumption values and future resale values. This modeling approach allows us to study the role of consumer expectation about future used-copy prices and resale values on current buying decisions. In particular, the used-copy inventory level, which affects the transition probability of resale values, plays a role of an exclusion restriction in the consumer selling decision problem to identify the discount factor. We estimate the proposed discrete choice dynamic programming model using the Bayesian algorithm by Ishihara and Ching (2016). To account for the potential endogeneity of prices, we apply the pseudo- policy function approach by Ching (2010b). The preference parameter estimates suggest that consumer heterogeneity in price sensitivity and costs of buying and selling used copies plays an important role in explaining the observed sales paths and substitution patterns. The demand patterns are well-explained by three types of consumers: (1) consumers who purchase new copies and become the main supplier of used copies in the used goods market (about 1% of the population), (2) consumers who purchase new and used copies, but are less likely to sell (about 25%), and (3) consumers who purchase new copies and sell, but their buying and selling probabilities are much lower than those for the first type of consumers, which explain the majority of non-purchasers (about 75%). The substitutability between new and used copies in the Japanese video game market highly depends on the second type of consumers. When the used-copy inventory is relatively low, new and used copies provide relatively similar utility for them as the price differential is canceled out by a high search cost for finding a used copy.

Load more